You may have me on that one actually, although as far as I know there's multiple antigens involved: insulin, GAD, and islet (I've not found any research as to what the specific characteristic they go after- if you find out, I'd be curious). Those are the core three I know of; it's worth noting that the point I raised also was raised by a few doctors in the referenced article (email 'em; might get a response where they clarify their views, or your question as to the antigen).
The closest paper I'm finding at this hour (fricking 5am my time) is their previous experiments w/ mice: http://diabetes.diabetesjournals.org/content/55/6/1546.full . If you find the dog paper, I'd be curious. The thing that's a bit weird there is a viral vector to induce an immune response for mice, *has* been accomplished before- rather than the cocktail (http://en.wikipedia.org/wiki/Streptozotocin) route they used. Presume they did the cocktail for control reasons, and since it's a helluva lot simpler.
Either way, the immune component of type1 isn't there in there experiment; as said, this is more like pancreatis induced type1. Doesn't mean that the research is crap, it just means that the immune interaction there is currently unknown
Re: the 'poo pooing'; keep in mind this is the NIH. Not to say they're perfect, but they're probably more trustworthy then some random jack ass on the internet- say me for example :) . I will say, their comments re: the speed of response to changing BG should be a concern that makes sense- the pancreas basically can dump it straight into the mainline of the body, muscles would have to effectively bleed it out meaning increased delays in addition to the sensing component potentially being delayed.
Either way, in my view, it's an interesting technique- but they've not tried it on a proper diabetic yet, just a disease model. Having it behave correctly despite a belligerent immune system would carry a lot of weight from my view.
That said, I still prefer the notion of just plain fixing the immune system mistargeting- hence the Faustman links. I'd strongly prefer to have my immune systems misbehaviour corrected- and pancreatic islet cell regeneration that occurs (hopefully to normal/full levels, although I've seen no studies that way)- rather than have a muscle mass converted into a quasi cgm/pump. Basically fix the cause, rather than go quasi artificial.
PS: In re-reading this, that's probably some of the worst english I've written yet. I blame the hour; pardon the wordiness. :/